American Cinematographer - March 2017 - 58

◗

Surveying the Virtual World

render each frame - and there's no
budging from that.
VR has been around for a really
long time, but we're finally at the
point where the hardware is available to
most people and can render creative,
interesting experiences within those 11
milliseconds. Coming out of visual
effects at ILM, xLab is a storytelling
division that's going for high fidelity. We
want everything to look as good as
possible within that timeframe.
Sean Safreed: Part of it is resolution and showing just what you need in
the current field of view. If you want a
real sense of immersion it needs to look
at least as good as HD. We're trying to
solve that with our adaptive streaming
- and in more than 4K. Soon we'll
move from mobile-derived displays to
next-generation displays. Instead of 1K
or 1,200 pixels per eye, you'll have 2K
resolution per eye. You really need 8K of
source resolution to fill the headset
display.
My sense down the road is we'll
see something spherical like the Nokia
Ozo [camera system], but instead of
having only a few cameras, you'll have a
lot of cameras on a spherical surface, and
you'll computationally stitch the final
image back together from all of these
58

March 2017

tiny sensors, which will give you
tremendously higher resolution. You'll
no longer be stuck with just a set of 2K
sensors each with a giant fisheye field of
view, because now you'll have dozens of
these sensors in a sphere. You'll be able
to compensate for the stitching errors
and parallax problems because there's so
much data overlapping in that field of
view. That's my prediction of what we'll
see in three to five years: a sphere
encrusted with little image sensors all
getting stitched within the device into a
perfect image with color and depth.
Ian Spohr: With VR, you're
changing rigs like you used to change
lenses, from studio modes to smaller,
more modular cameras when you need
the mobility. You're also fighting the
laws of physics. Digital cinema is catching up to film, where a bigger sensor
means better image quality, but in VR
we need smaller systems that can
achieve that same image quality. Just to
be able to look around with a good
composition and dramatic elements
means we're shooting VR with camera
systems that range from the size of a
softball to the size of a go-kart.
Michael Mansouri: A major
limitation is not just from the camera
side but the headset technology. You
American Cinematographer

actually feel the softness, or 'screen-door
effect,' in the image, which is due to
looking into a diopter at a mobile device
screen and revealing around 9 pixels per
degree of resolution. A motion-picture
film scan is up to 50 pixels per degree,
while the best VR headset on the
market today is about 14 pixels per
degree. In order for us to have the same
visual sharpness and clarity, we need to
have from 24 to 27 pixels per degree.
We're not there yet, but we're going to
get there; it's just a matter of time.
Postproduction is also a massive
limitation. That's improving with a lot
of amazing technology to speed up and
automate that process, so not every
frame requires a visual-effects fix. New
technology will also enable filmmakers
to upload VR content into a cloud to
computationally stitch without a lot
of manual labor.
Malte
Wagener:
The
challenge on the interactive VR side is
technological, and with 360-degree
video it's more intellectual. For 360, how
do you keep the audience engaged after
their first, second and third 360 experiences? You're showing people a certain
perspective for two or three minutes for
most 360 experiences; how you tell a
story is an interesting problem to solve.

Location photo by Marcus Johannes, courtesy of Radiant Images. Callahan photo by David Stump, ASC.

Above: The Dark Corner rig, fitted with Sony a7S II cameras, on location in Tasmania for a two-week
commercial shoot. Right: Stump's assistant Lauren Callahan sports the Mobius POV VR 360 rig.

Table of Contents for the Digital Edition of American Cinematographer - March 2017